Deliverable No 2.3. Inventory of Available Information

Transcription

1 EUROPEAN COMMISSION - EUROSTAT ESSnet on consistency of concepts and methods of business-related statistics 2010 project on statistical units n Deliverable No 2.3 Inventory of Available Information Author: Joseph Keating - CSO February 2012

2 Contents 1. Introduction Summary of Findings Group A Regulations (Legal acts providing a methodological/conceptual basis for surveys and data compilation)... 8 External Study... 8 Eurostat workshop... 9 Dublin workshop Group B Regulations (Legal Acts governing surveys, data collection and statistical indicators) External Study Eurostat workshop Dublin workshop Group C Regulations (Legal Acts referring to statistical projects which might be characterised as Tertiary statistics ) External Study Eurostat workshop Dublin workshop Enterprises and Legal Units External Study Eurostat workshop Dublin Workshop Kind of Activity Unit(KAU)\Local KAU\Local Unit External Study Eurostat workshop Dublin workshop... 22

3 8. Enterprise and Enterprise Group External Study Eurostat workshop Dublin workshop Special Purpose Enterprises (SPEs) External Study Eurostat workshop Dublin workshop Appendix -Coverage of statistical domains... 27

4 1. Introduction This document describes work done to date (December 2011) on the ESSnet on Consistency of Concepts and Methods of business- and trade-related Statistics Work Package 1(Statistical Units). It summarises the information about Statistical Units that has been collected so far. The ESSnet sent a questionnaire on Statistical Units to the member states in late 2011, but the results of the questionnaire are not included in this inventory. The ESSnet on Consistency Work Package 1(WP1) focuses on Statistical Units. It began in January 2011 and will end in December The ESSnet is being co-ordinated by Italy (ISTAT), and four other National Statistical Institutes (NSI s) are participating, Austria, France, Ireland and Latvia. There are 2 other work packages in the ESSnet on consistency. These are Work Package 2(WP2) on Target populations, frames, reference period, classifications and their applications (WP2) and Work package 3(WP3) on Characteristics and definitions. WP2 will start in January 2012 and WP3 later in The ESSnet on Consistency falls under action 2.1 (integrating concepts and methods within the legal framework) of objective 2 (achievement of a streamlined framework for business related statistics) of the MEETS programme. More detail about the context and proposed direction of WP1 can be found in the linked Direction Report. ( Deliverable 1.1 Direction report) This document focuses on the information about statistical units collected from 3 sources. 1. The external study by Josef Richter, Christian Engelage, and Peter Thomas entitled External study on detail evaluation of the legal acts in the areas of statistics which were identified by member states as areas to revision. This was submitted to Eurostat in July It consists of a description of all the regulations which were deemed relevant to the Consistency ESSnet(see Appendix) and a description of how consistent or inconsistent they are with respect to each other. The study focuses on all aspects of consistency among the regulations, but in this document we will only highlight parts of the study that refer to the use of statistical units in the various regulations. This study can be found at: 2. A Eurostat workshop held with Eurostat experts in Luxembourg held on 2 nd and 3 rd March It was the aim of this workshop to gather all relevant information from the Eurostat experts responsible for the various statistical domains which are to be considered in this WP1. 3. A Dublin workshop with national experts held on 12 th and 13 th April This workshop consisted of a series of presentations from experts from various NSI s. Minutes of the Eurostat and Dublin workshop can be found at: (Deliverables 2.1 and 2.2)

5 This document is structured as follows: We begin with a brief summary of the main findings. We then present the information learned from the 3 sources mentioned above by Groups of Regulations. The groupings used are those identified by Dr. Richter and his colleagues in their External Study. These are (see Appendix); Group A Regulations (Legal acts providing a methodological/conceptual basis for surveys and data compilation) Group B Regulations (Legal Acts governing surveys, data collection and statistical indicators) Group C Regulations (Legal Acts referring to statistical projects which might be characterised as Tertiary statistics ) Next we present the information learned from the 3 sources mentioned above by issue. The issues are: Enterprises and Legal Units Kind of Activity Unit(KAU)/Local KAU/Local Unit Enterprise and Enterprise Group Special Purpose Enterprises By presenting the information in this composite manner we aim to give as comprehensive a view as possible of the information gathered to date. Direct quotes from the External Study or from the minutes of the workshops are indicated in italics.

6 2. Summary of Findings The ESSnet Work Package 1 on Consistency of Statistical Units covers a lot of ground. 18 Regulations were looked at. There are 4 core Statistical Units looked at in the project (Enterprise, Kind of Activity Unit (KAU), Local Unit, Local KAU), as well as units unique to certain regulations. Finally, the practices of all the MS in implementing these regulations with regard to all the statistical units have to be considered. This presents a complex picture and it can be difficult to see the crucial issues. However, we can identify a spine of the most important regulations with regard to statistical units, and see that there are major inconsistencies even here. This spine consists of the Statistical Unit Regulation and Business Register Regulation in Group A (see Appendix for groupings), the SBS and STS Regulations in Group B, and the National and Regional Accounts Regulation in Group C. The inconsistency of the current regulations can be seen in the diagram below. The Statistical Unit Regulation distinguishes and defines Enterprise, KAU, Local Unit, Local KAU (and other units). The Business Register Regulation requires information to be held on Enterprise and Local Unit but not on KAU or Local KAU. Some variables in the SBS Industry and Construction modules should be based on KAU as well as on Enterprise, some on Enterprise only, The other SBS modules are based on Enterprise (modules on Insurance and credit institutions an exception). The Industry and Construction Annexes of the STS Regulation specify KAU as the only statistical unit to be used. The Retail and Other Services Annexes specify Enterprise. The National Accounts Regulation specifies Institutional Units (equivalent to Enterprise in business related statistics) as the unit for describing income, expenditure and financial flows, and balance sheets. It specifies Local KAU as the unit for describing processes of production and for input-output analysis.

7 The situation is further complicated when we look at other regulations, some of whom use units as defined in the statistical unit regulation, some of whom define their own statistical units, and in some the statistical unit to be used is undefined. When we look at the practices of member states in implementing these regulations, it is clear from both workshops that a large number of member states find the current definition of Enterprise difficult to implement, and settle for an identification between legal unit and Enterprise, ignoring the criterion of autonomy. MS who do aim to satisfy the current definition tend to find the profiling necessary a resource intensive task. Similarly the profiling necessary to monitor Enterprise Groups is resource intensive. The implementation of KAU and LKAU in different MS and in different Regulations appears to be quite diverse, with many different approaches taken. Some MS have KAU on their Business Register, but many do not (following the BR Regulation). Some MS use Local Unit as a proxy for KAU, and some use Enterprise. The need for a satisfactory and practical definition of Enterprise Group is particularly needed in those statistical areas with a strong International/Global Dimension, such as FATS, Foreign Direct Investment statistics, and Balance of Payments. The work of the Profiling ESSnet (see pages 24-25) suggests an international approach to this issue. If this approach were adopted, it would also enable MS to have better defined national Enterprises for complex Multi National Enterprises. Some MS are finding it necessary to reliably identify Special Purpose Enterprises (SPE s) The current regulations give little or no guidance in how to approach this issue. It is difficult to quantify how much and in what manner all these issues in the area of Consistency of Statistical Units affect the statistics produced. For some variables (e.g. employment) the overall totals should not be affected, but will be classified to the wrong NACE code, region, country etc. For other types of variables, inconsistency of statistical unit will effect the totals themselves (see the example of Sold Production on page 19). The French examples on page 25 indicate that in some instances the inconsistency can be large.

8 3. Group A Regulations (Legal acts providing a methodological/conceptual basis for surveys and data compilation) External Study There are 2 regulations in this category that are of fundamental importance in discussing the issue of consistency in statistical units Regulation No. 177/2008 Business Register Regulation Regulation No. 696/93 Statistical Units Regulation The Statistical Units regulation distinguishes and defines eight statistical units: (see External Study pages 22 to 24) 1. Enterprise the smallest combination of legal units that is an organisational unit producing goods or services, which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. 2. Institutional Unit For the purpose of Business-Related statistics, the regulation states that In the corporate Enterprises sector, the Enterprise corresponds to the institutional unit used in the ESA. So for our purposes, Enterprise and Institutional Unit can be treated as synonyms. (Institutional Unit can mean a household in non business-related statistics, but this is not relevant for our purposes). 3. Enterprise Group An association of Enterprises bound together by legal and/or financial links. A group of Enterprises can have more than one decision-making centre, especially for policy on production, sales and profits. 4. Kind of Activity Unit (KAU) Groups all the parts of the Enterprise contributing to the performance of an activity at Class level (4-digit) of NACE. The KAU must be an observable unit. To achieve this, KAU s might not be perfectly homogenous with respect to their activity. 5. Unit of homogenous production Same idea as KAU, but strictly homogenous, and thus not necessarily observable 6. Local Unit An Enterprise or part thereof situated in a geographically identified place. 7. Local Kind of activity unit (LKAU) the part of a KAU which corresponds to a local unit. 8. Local Unit of homogenous production Same idea as LKAU, but strictly homogenous, and thus not necessarily observable. As Institutional Unit is equivalent to Enterprise in business related statistics, and Units of homogenous production are more conceptual than practical. The ESSnet on EGR is dealing with concepts and definitions of Enterprise Groups (see ). In this ESSnet Enterprise Groups will be considered as surroundings in which to analyse the other statistical units. Thus, in this project we focus on the 4 units Enterprise, KAU, Local Unit, and LKAU. Turning to the Business Register Regulation, only 3 of these units are explicit requirements, Enterprise, Enterprise Group and Local Unit. The Business Register Regulation does not explicitly require KAU and Local KAU. However, the regulation does require that among the variables to be recorded in the Business Register for both Enterprises and Local Units are: Principal activity code at NACE 4-digit level and Secondary activities, if any at NACE 4-digit level. This knowledge of secondary activities could help to determine KAU s and LKAU s, if they are not recorded in the Business Register as separate units.

9 Regulation 1893/2006 (Classification of Economic Activities NACE Rev. 2) states that statistical units referred to in business registers, as set up according to Regulation No 2186/93, shall be classified according to NACE Rev. 2 (page 30) The other Group A regulations do not have any issues with regard to statistical units. Eurostat workshop A presentation was given by Eurostat Business Register experts. They stated that the business register should promote a homogenous use of statistical units by all the statistical domains and is very much interested in the potential changes of definition of the statistical units. They also highlighted the EuroGroups Register (EGR) and stated that the role of the EGR should be emphasized in the future to improve the coordination between countries and the homogeneity of statistics for statistical units in the frame of multinational Enterprise groups Dublin workshop Presentations from Business register experts from Denmark, Ireland, Finland, Slovenia and Switzerland were made at the Dublin workshop. These gave a picture of how in some cases countries practices differ from the Business Register or Statistical Unit Regulation. For example: The statistical units on the Finnish Business Register are: Legal Unit, Local Unit, Enterprise Group and LKAU. Enterprises are not available on the Business Register. However a project is currently underway to develop a Business Data Warehouse that will be able to provide all statistical units required under regulation. The statistical units will include units at Enterprise and KAU level. The statistical units on the Slovenian Business Register are Legal Unit, Administrative Local Units and branches of foreign owned business entities. In the case of both countries, the statistical units on the register are a consequence of the administrative sources that the respective Business Registers rely on. Switzerland is an example of a Business Register that fully implements the current requirements of the Business Register regulation, and has no plans to introduce extra units such as KAU or LKAU.

10 4. Group B Regulations (Legal Acts governing surveys, data collection and statistical indicators) External Study The external study identified four different levels of consistency between the many regulations in this category and the Statistical Unit regulation. 1. Units in the regulation are as defined in Statistical Unit Regulation, and there are explicit references made to the Statistical Unit Regulation. Examples of Regulations with this approach are the Structural Business Statistics (SBS) regulations and Short-Term Statistics (STS) regulations. 2. Units in the regulation are as defined in Statistical Unit Regulation, but there are exceptions, and the link is rather weak. An example of a regulation with this approach is the Science and Technology Statistics Regulation 3. Units in the regulation are well-defined but do not refer to the Statistical Unit Regulation, or only in a very indirect way. An example of a regulation with this approach is the Intrastat Regulation. 4. The units in the regulation are more or less undefined. An example of a regulation with this approach is the Job Vacancy Statistics Regulation. See Table 2 on pages of the External Study. Other inconsistencies considered to be of central relevance in the report are: In SBS Module V (Insurance) also relies on businesses, in Module VI credit institutions are addressed. In the Common Module (Module I) as regards the regional statistics covered in section 4.5 no statistical unit is specified. In Prodcom other statistical units than Enterprises may be used as observation units as long as Enterprise data is transmitted to Eurostat. In Energy Statistics the statistical unit is not explicitly defined. In Waste Statistics either local units or KAUs may be used For R&D Statistics no definitions of the statistical unit to be used are given in the legal acts. Labour Cost Statistics no specific provision as regards the statistical unit is given. Labour Cost Indices reference is made to Regulation No 696/93, but there is no provision which of the units defined in this Regulation has to be used. (page 265) The report (see page 197, figure 8) also classifies the many Group B regulations into the following clusters. The clusters are classified both by whether they are oriented towards Economic Units, Persons, or Commodities, and by whether they are Short Term or Structural. Different parts of certain regulations occur in more than one cluster:

11 Economic Units/Activity oriented Person/household oriented Short Term Structural Analysis Short Term Statistics (STS) Structural Business Statistics (SBS) Tourism FATS Job Vacancy statistics Environmental expenditure Labour Cost Index Statistics on Information and Communications Technology Research and Development Structure of Earnings Survey/Labour Cost Survey Vocational Training statistics Foreign Direct Investment International Trade in Services Labour Force Survey Tourism Statistics on Information and Communications Technology Structure of Earnings Survey/Labour Cost Survey Commodity Oriented (page 197) Energy External Trade Prodcom Energy Waste statistics External Trade As might be expected, the majority of Consistency issues relating to Statistical Units occur in the Economic Units/Activity oriented Regulations, so we concentrate on these: Economic Units/Activity oriented Regulations With regard to short-term regulations in this category the report states: The main inconsistency must be seen in the different statistical units on which the closely related projects rely: STS are based on KAUs in Industry and Construction but on Enterprises in Retail Trade and Other Services. In particular when administrative data are to be used for the compilation it will be difficult to avoid double counting and overlapping such as in the case of manufacturing activities carried out in separate KAUs with that by Enterprises classified outside Industry and Construction. Tourism Statistics are based on LKAUs. Job Vacancy Statistics rely on business units which are not defined in terms of Regulation No 696/93 on the statistical units for the observation and analysis. The underlying observational unit of Labour Cost Index is also not defined. (page 259) With regard to Structural Analysis regulations in this category the report states that SBS is the central regulation, and that the other regulations in this category may be seen as satellites of SBS. It then sums up the use of statistical units in this area as follows:

12 The Enterprise is the dominating statistical unit in this field of closely related statistical projects. Additional information on the KAU level is collected in a number of areas as part of SBS. FATS uses branches as additional statistical units. In Statistics on R&D the statistical unit is strictly speaking not explicitly defined, although the Frascati Manual recommends using with some exceptions the Enterprise unit as the reporting unit in the business Enterprise sector. The Structure of Earnings Survey and the Labour Cost Survey are to be based on local units and Enterprises. (page 261) Finally, we focus specifically on the two main Economic Units/Activity oriented Regulations, STS and SBS. The units required by these regulations are summarized in the following tables: (based on tables page ) SBS Module Statistical Units NACE Coverage Comment required I Enterprise 05-82, 95 Common Module II Enterprise, KAU Industry III Enterprise Trade IV Enterprise, KAU Construction V Business Insurance VI Enterprise, Branches 64.19, Credit Institutions VII Enterprise, Ancillary 65.3 Pension Funds Activities VIII Enterprise 58.2, 62, 63.1, 69.1, Business services 70.2, 71.1, 73.1, 78 IX Enterprise Business Demography Some of the modules in SBS require different variables to use different statistical units. The Industry module requires variables such as turnover, production value, wages and salaries, gross investment in tangible goods and number of persons employed to be compiled at KAU level, as well as at Enterprise level, while other variables need only be supplied at Enterprise level. A similar situation occurs in the Construction module.

13 STS Annex Statistical Units NACE Coverage Comment required A KAU 05-33, , 35- Industry 36, 38.3 B KAU Construction C Enterprise 47 Retail trade and repair D Enterprise 45-46, 49-82, Other Services It can be seen from these tables that both SBS and STS use statistical units defined in the Statistical Unit Regulation (with some exceptions in the case of SBS). However, it can also be seen that both regulations require reporting by KAU, which is not a statistical unit governed by the Business Register regulation. Eurostat workshop There were 17 presentations from experts on different Group B regulations at the Eurostat workshop. We group them below using the schema in the table on page 11 above. Economic Units/ Activity oriented (Short-Term) survey Regulations Presentations were given in this cluster concerning the Short Term Statistics, Labour Cost Index, Job Vacancy and Tourism Regulations. The STS expert discussed how the STS Regulation specifies the observation unit for each annex: KAU for Industry and Construction, and Enterprise for Retail Sales and other Services. He also noted that for Industry and Construction, instead of KAU, the local unit or Enterprise may be used for Enterprises with few persons employed in secondary activities. Some MS have difficulties in collecting information on KAU s and use this rule widely, with some MS collecting data in Industry and Construction by Enterprise, and some others using local unit. For Labour Cost statistics, the majority of the member states collect data for the local unit, but 7 MS collect data for the Enterprise, causing comparability problems. For Job Vacancy statistics, the statistical unit according to the regulation is the Business unit. As noted above, this does not refer to the Statistical Unit Regulation, and is more or less undefined. The expert presenting said that in practice, the majority of reporting countries use Enterprise as the statistical unit, but 8 member states collect data from local units. Using two different statistical units causes a problem of comparability particularly in terms of NACE classification.

14 For Tourism the expert stated that the statistical unit is the LKAU or the local unit. He also stated that information on statistical units relevant to Tourism is not well covered by business registers and therefore many MS operate a specific register for Tourism. Economic Units/ Activity oriented (Structural) survey Regulations Presentations were given in this cluster concerning the Structural Business Statistics, Business Demography, FATS, Direct Investment, Structure of Earnings, R&D, Community Survey on Innovation, and Community Survey on Information Society Regulations. The SBS expert s presentation included some information on how MS implement the statistical units specified in the SBS Regulation (the Enterprise, Local Unit, and KAU) in practice. This will be discussed in more detail in the sections below. In general, it is clear that in several countries the definitions of Council Regulation (EEC) No 696/93 of 15 March 1993 on the statistical units for the observation and analysis of the production system in the Community are not yet fully respected. The statistical unit in Business Demography is the Enterprise. The statistical units used in FATS are the Enterprise and the branch, but the definition of branch is weak and information on branches is often hardly available. For Direct Investment Statistics, The expert stated that there are two approaches to the definition of the statistical unit used in this domain. Both approaches are accepted in the manuals and are used within the EU. One approach uses the Institutional Unit (IU) as the statistical unit, which is applied in 23 MS. The other approach uses the Local Enterprise Group (LEG), which is applied in the rest of the MS. The expert also mentioned the importance of improving consistency with other institutions such as the OECD and the UNO. For Structure of Earnings statistics, the majority of the MS collect data for the local unit but 8 MS (mainly small MS) collect data for the Enterprise. This can cause some problems according to comparability. The R&D expert said that the Enterprise is the unit used in R&D statistics. While no definition of unit is given in the Regulation, a reference is given to the methodological framework of the Frascati Manual that recommends the Enterprise as statistical unit for the domain. An exactly similar situation occurs in Community Survey on Innovation statistics, the reference in this case is to the Oslo Manual. The Community survey on Information Society regulation specifies the Enterprise as the statistical unit to be used.

15 Person/household oriented survey Regulations The only presentation specific to this area given at the workshop was on Labour Force statistics. Because the reporting unit is the individual person there are no known issues with Statistical Units. Commodity oriented survey Regulations There were presentations in this area from experts on PRODCOM, External Trade, Energy and Waste Statistics. The observation unit in PRODCOM is the undertaking (= Enterprise). Inconsistencies arise from how MS build up the Enterprise from Legal units. In External Trade Statistical Units have not been an issue up to recently. The reporting units are traders identified by VAT number and/or Customs number. However a new Regulation (NB not covered by the External Study) includes a module on Trade by Enterprise characteristics. These statistics are to be compiled by merging trade micro-data with Business Register information. The Energy expert stated that the notion of the statistical unit used in this domain is not well defined. Different observation and reporting units are used. There is no clear documentation in detail of how the MS apply the statistical unit. There is a wish to improve consistency with non-eu Energy statistics. In Waste Statistics 10 MS use the local unit and 3 MS use KAU (other MS have not reported yet). The expert stated that the effect of choice of the statistical unit to the quality of the domain is not known but is expected as very low because of the fact that not the number of units is reported but tonnes of waste by NACE activity at highly aggregated level. Dublin workshop Presentations from an SBS expert (from Netherlands) and an STS expert (from Italy) were given at the Dublin workshop. The Netherlands expert described how a redesign in the way statistical units in the Netherlands are constructed has led to (a) more use of administrative data (b) more methodological coherence and (c) more streamlined processes but with the disadvantages of (a) more heterogeneity and (b) difficulties arise for respondents answering. The Italian expert highlighted the traditional asymmetry between industry and services in the STS regulation, in particular the greater detail required of industry. This point was also highlighted by the Finnish Business Register expert who pointed out that the inconsistency within the STS Annexes can lead to activities being missed from all Annexes, if the Regulation is followed strictly. In particular, a Service KAU within an Industry Enterprise will not be reported in the Industry Annexe (because the KAU is not Industry), but will also not be reported within Services (because the Enterprise is not Services).

16 5. Group C Regulations (Legal Acts referring to statistical projects which might be characterised as Tertiary statistics ) External Study The European System of National and Regional Accounts (ESA) uses two types of Statistical Units for different analytical purposes. In order to describe income, expenditure and financial flows, and balance sheets, the system groups institutional units into sectors on the basis of their principal functions, behaviour and objectives. In order to describe processes of production and for input-output analysis, the system groups LKAUs into industries on the basis of their type of activity. (page 185) As explained earlier, the Institutional Unit can be seen as equivalent to the Enterprise in business related statistics. The LKAU as defined in the ESA is compatible with the definition in the Statistical Unit Regulation. But there is no requirement for LKAU in the Business Register Regulation. Neither does the ESA mention the Business Register for statistical purposes. Furthermore, none of the most important Regulations for surveys on which National and Regional Accounts are dependent specify that information should be collected by LKAU, for example, SBS, STS, Prodcom, External Trade Statistics. The role for providing one of the main sources for compiling National Accounts is not mentioned among the purpose of SBS. (page 39) The Balance of Payments (BOP) Regulation includes no explicit provision as regards statistical units (page174) and makes no reference to the Statistical Unit Regulation. Eurostat workshop A presentation on National Accounts was given. The expert stated that the MS do not apply the statistical units in the same way, and use different time intervals (annual or continuous) to update the LKAU. Dublin workshop Presentations from a National Accounts expert (from United Kingdom) and a Regional Accounts expert (from Austria) were given at the Dublin workshop. The Regional Accounts expert highlighted that: A characteristic feature of the system is the use of two types of unit and two ways of subdividing the economy, which are quite different and serve separate analytical purposes. In order to describe income, expenditure and financial flows, and balance sheets, the system groups institutional units into sectors on the basis of their principal functions, behaviour and objectives.

17 In order to describe processes of production and for input-output analysis, the system groups local kind-of-activity units (local KAUs) into industries on the basis of their type of activity. An activity is characterised by an input of products, a production process and an output of products. (ESA 1.27). She also analysed the consequences for Regional Accounts of not having KAU (discussed below). The UK expert stressed the importance of a single, comprehensive, high-quality Business Register to produce high quality National Accounts, and stated that this implies the Business Register working at the lowest level of detail, i.e. at LKAU.

18 6. Enterprises and Legal Units External Study The Statistical Unit Regulation defines an Enterprise as the smallest combination of legal units that is an organisational unit producing goods or services, which benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. (page 22) The Business Register regulation requires that information on both legal units and Enterprises are held on the Register. Local Units, KAUs and LKAUs are derived from Enterprises. Thus the method by which countries combine legal units to form Enterprises, following the definition above, is of great interest from the point of view of consistency. Eurostat workshop The experts on the Business Register stated that the MS may have difficulties to apply definition of the Enterprises that are in the SU regulation because in their understanding the definition seems to be too theoretical and should be more pragmatic They showed the results of a 2006 survey of NSI s, showing that: 9 NSI s had a BR that had only legal units without links between legal units 3 had only legal units with links between legal units 2 had only statistical units comprising several legal units 3 had legal units and Enterprises (i.e. what the BR Regulation specifies) 6 had legal units and groups In the same survey, MS said that the most common unit used to perform SBS statistics was the legal unit. More evidence comes from the EuroGroups Register (EGR). If one looks at the average number of legal units per Enterprise in the different reporting countries, we see that only 2 countries (Netherlands and UK) frequently define Enterprises as a set of legal units. Netherlands has an average of over 2 legal units per Enterprise in the EGR, and UK has an average of over 1.5 legal units per Enterprise. All other MS have an average of exactly 1(i.e. legal unit = Enterprise always) or very slightly greater than 1. The experts also stated that some countries have experimented the need to split legal units into several Enterprises in the case of functional (or transversal) activities that are run at the global level of the group and that serve several Enterprises. Note that the current definition of Enterprise does not allow this type of split. Many of the other experts highlighted as a problem that many MS do not follow the definition of Enterprise as a combination of legal units. For example:

19 FATS: most MS follow the concept of defining an Enterprise as only one legal unit and not as a combination of legal units as stated in the Statistical Unit Regulation Business Demography: Most MS use in fact the legal unit as statistical unit because they do not create complex Enterprises. This is done because most of the (administrative) data are available only for the legal unit and not for the Enterprise which is an artificial statistical unit and has no direct linking to administrative data without profiling. SBS : The number of MS building up Enterprises is growing constantly but profiling is much effort and in many cases only big Enterprises are established which leads to problems in comparability of data. Prodcom: How MS build up and define the unit Enterprise can effect the variable Sold Production. There is a danger that some of the declared Sold Production could include sales to another legal unit that should be included in the same Enterprise. If some MS just have a 1 to 1 relationship between legal unit and Enterprise, there is a risk of over-estimating Sold Production. Dublin Workshop Experts from Denmark, Ireland, Netherlands, and UK gave presentations in the session on Enterprises and Legal Units at the Dublin workshop. Denmark follows the Eurostat definition of Enterprise where it is practical to do so: 35 complex Enterprises have been profiled and it takes at least two days to profile one and seven days to combine information from units. In the case of smaller companies no statistical advantage is seen in strictly applying the Eurostat definition of Enterprise. Difficult cases include holding companies, Building & Construction, Public Utilities, Public Sector, agricultural holdings. The UK also highlighted that the profiling necessary to turn legal units into Enterprises is resourceintensive. UK indicated that they have 4-5 staff profiling full time a dedicated resource in this regard is important. In Ireland, Some historical cases exist with several legal units on one Enterprise, but in general the one to one link between legal units and Enterprises is considered the most useful. However this can lead to some inconsistencies. For example, an accountancy partnership filing 1 VAT return would be treated as one legal unit and 1 Enterprise. However, if the partners file VAT returns separately, they will be treated as separate legal units and separate Enterprises. Similar issues occur in the health and education sectors.

20 7. Kind of Activity Unit(KAU)\Local KAU\Local Unit External Study The role of these units in the Eurostat Regulations is as follows: Kind of Activity Unit (KAU) The Statistical Unit regulation specifies KAU as follows: The KAU groups all the parts of an Enterprise contributing to the performance of an activity at Class level (4-digit) of NACE, and corresponds to one or more operational subdivisions of the Enterprise. The concept behind identifying a KAU is therefore to arrive at units with a high degree of homogeneity with respect to the activity, but taking into account that KAUs are not perfectly homogenous with respect to their activity (page23) and the Enterprise s information system must be capable of indicating or calculating for each KAU at least the value of production, intermediate consumption, manpower costs, the operating surplus and employment and gross fixed capital formation.( page24) The Business Register Regulation does not require BR s to include KAU s. It does require that Secondary activities, if any, at NACE 4-digit level be among the variables of the Enterprise units held on the BR. This characteristic offers the possibility of analytically defining KAU. As described above, some variables for some modules in SBS are required to be reported using KAU as the observed statistical unit. The Industry and Construction Annexes in the STS regulation also require using KAU. None of the other survey(group B) regulations require using KAU as the statistical unit. The National Accounts Regulation specifies LKAU (which implies a knowledge of KAU) as the statistical unit in order to describe processes of production and for input-output analysis. Local Unit The statistical unit Regulation defines a local unit as An Enterprise or part thereof situated in a geographically identified place (page 24) The Business Register Regulation requires BR s to include Local Units. The Structure of Earnings Survey/Labour Cost survey is the only survey (Group B) regulation requiring Local Unit as the statistical unit. (Table ) The National Accounts Regulation specifies LKAU (which implies knowledge of Local Unit) as the statistical unit in order to describe processes of production and for input-output analysis.

21 Local KAU The Statistical Unit Regulation defines a LKAU as the part of a KAU which corresponds to a local unit. It is the observational unit with a maximum of homogeneity as far as activity and region is concerned (page 24) The Business Register Regulation does not require BR s to include LKAU s. It does require that Secondary activities, if any, at NACE 4-digit level be among the variables of the Local units held on the BR. This characteristic offers the possibility of analytically defining LKAU. Survey (Group B) Regulations requiring LKAU as the statistical unit are Tourism and Waste Statistics. (Table ) The National Accounts Regulation specifies LKAU as the statistical unit in order to describe processes of production and for input-output analysis. Eurostat workshop The SBS expert presented a table showing the ratio of number of KAUs to the number of Enterprises for each MS for NACE codes C, D, and E (2004 data). We would expect this ratio to be greater than 1 as each Enterprise has at least 1 KAU, and some will have more. The table shows 7 MS who do indeed have a ratio of greater than 1, the highest ratio being Germany with a ratio of MS have a ratio of exactly 1, indicating that Enterprise and KAU always have a 1 to 1 relationship in these MS. 4 MS have a ratio of less than 1, which could indicate a size class threshold for the KAU statistics. A similar table was presented for the ratio of number of Local Units to the number of Enterprises for each MS for NACE codes C, D, and E Again we would expect the ratio to be greater than 1. This is in fact the case for 19 MS, with the highest ratio being 1.34 for Latvia. The ratio equals 1 for 2 MS and is less than 1 for 3 MS. The STS expert discussed how the STS Regulation specifies the observation unit for each annex: KAU for Industry and Construction, and Enterprise for Retail Sales and other Services. He also noted that for Industry and Construction, instead of KAU, the local unit or Enterprise may be used for Enterprises with few persons employed in secondary activities. Some MS have difficulties in collecting information on KAU s and use this rule widely, with some MS collecting data in Industry and Construction by Enterprise, and some others using local unit. The expert stated that the STS guidelines justify the use of KAU for the following 2 reasons: "The KAU is meant to reduce the heterogeneity according to activity, which is inherent to the Enterprise. At the same time, it tries to avoid being an artificial construct that could not be implemented." This aspect is considered to be important especially for (quarterly) national accounts where STS indicators (above all the index of industrial production and the index of industrial output prices) are used as input. The STS guidelines, still provide another reason for the choice of the KAU as observation unit in STS: "The KAU was devised as observation unit in STS in order to improve the homogeneity of the results of

22 statistical surveys by activity and hence the international comparability of these results, since at the level of the Enterprise different types of horizontal and vertical integration can be observed at both national and international level." This aspect is of particular importance if we want STS to be able to follow economic trends like the move away from "traditional" (manufacturing) activities towards service activities, a move which can be observed both at the level of the (big) Enterprises and of national economies. For the Retail and other Services annexes, the reliance on Enterprise can be limiting where Enterprises have highly heterogeous activities. A good example being the major chains of petrol stations which can be classifies as food catering Enterprises. Use of KAU would give a clearer picture of what businesses are actually doing in these cases. Another consideration is whether the further integration of STS with other areas of business statistics (particularly SBS) requires a change in the observation unit used in STS. Dublin workshop Experts from Italy, Finland, Slovenia, Switzerland and Austria gave presentations in the session on Kind of Activity Unit(KAU)\Local KAU\Local Unit at the Dublin workshop. The statistical units on the Finnish Business Register are; Legal Unit, Local Unit, Enterprise Group and LKAU. Enterprises are not available on the Business Register. However a project is currently underway to develop a Business Data Warehouse that will be able to provide all statistical units required under regulation. The statistical units will include units at Enterprise and KAU level. Many Legal Units in Finland find it difficult to report at LKAU level. The focus on reporting at LKAU is on the Industry sector. In the Services sector, Data is technically generated to LKAUs based on Legal Unit information. The Swiss BR follows the BR Regulation, and does not have KAU or LKAU, and has no plans to add them to the register. However, the information stored on certain Enterprises in the register is detailed enough that it can enable the working out of a virtual KAU or virtual LKAU which can then be offered to National Accounts. The Enterprises for which this is available are those for which profiling has been done, or for multi-establishment Enterprises. These make up 5% of Enterprises and 40% of employment. KAU is also not kept as a separate unit on the Slovenian BR. but 44% of Legal Units have registered secondary activities. The register also holds Administrative Local Units (ALU) which are subsidiaries of Legal Units that can perform different activities as legal units on the same location as the Legal Unit or the same or different activities at an addresses other than the Legal Unit. In Italy, SBS and STS monitor KAU using their own registers, updated monthly in the case of the STS register, and annually in the case of the SBS register. The Italian expert also presented a table of results showing Industry turnover by NACE, and how this can differ greatly at NACE Division level, depending on whether KAU is the statistical unit used or not. The Austrian expert on Regional Accounts described how gross value added (GVA) is calculated.

23 85% of GVA in regional accounts is used by compiling data using a bottom up (or pseudo bottom up) approach from LKAU or KAU data. Three approaches based on the relationship between the Enterprise (or multi-establishment Enterprise) and the LKAU were outlined. 50% were based on a bottom up approach where each Enterprise was equivalent to a KAU which in turn was equivalent to a LU. 35% were based on a pseudo bottom up approach where Enterprise comprised of one KAU. However in the pseudo bottom approach the KAU was divided among various LUs (LKAUs) often located in different regions. The remaining 15% involved more complex multi-establishment Enterprises comprising of a number of KAUs (often split into a number of regions) and in these cases a top down approach was adopted. The expert also said that Losing information at KAU level would lead to weaker assumptions for the industry level of measures in national and regional accounts, less accurate regional data, less accurate production data for industries, and less meaningful analysis of productivity

24 8. Enterprise and Enterprise Group External Study The Statistical Unit Regulation defines an Enterprise Group as an association of Enterprises bound together by legal and/or financial links. A group of Enterprises can have more than one decision-making centre, especially for policy on production, sales and profits. (page 23) The explanatory notes for the regulation point out that the Enterprise group unit although particularly useful for financial analyses and for studying company strategies is too unstable to be adopted as the central unit for observation and analysis. (page 23) The Business Register regulation requires that information on Enterprise Groups is held on the Register. Having information on Enterprise Groups is of particular relevance to those surveys that have an International/Global Dimension, such as FATS, Foreign Direct Investment statistics, and Balance of Payments. Eurostat workshop The expert on Business Register and EuroGroups Register commented on the work of the ESSnet on Profiling, and their proposal to define Enterprises within large and complex Multi-National Enterprises at 2 levels: the global level of the group and the national level. The expert stated that the current situation where Enterprise is only defined at a national level leads to discrepancies and irrelevant statistics in the context of globalization and organization of the groups at the global level. However, there is still a need for national statistics, so the national level is still required. Moreover, economic studies on the productive system usually don't take into account the international dimension of the Enterprises. This can lead to wrong interpretations. The expert on Direct Investment Statistics commented that 4 MS use Local Enterprise Group as their statistical unit. The R&D expert stated that a small number of countries use Enterprise Group (instead of Enterprise) as their statistical unit. Dublin workshop Experts from Netherlands, France and Ireland gave presentations in the session on Enterprise and Enterprise Group at the Dublin workshop. The Netherlands expert described the work of the ESSnet on Profiling multinational and complex Enterprise groups. See The expert defined profiling as a method to analyse legal, operational accounting structures of a group in order to establish the statistical units within that group, their links and the most efficient structures for the collection of statistical data. The ESSnet on Profiling is proposing changes to the definitions of Enterprise Group and Enterprise, and 2 additional units, Truncated Enterprise Group and Truncated Enterprise (the position and definition of the

25 KAU is to be discussed). Their proposal is that instead of combining legal units to form Enterprises (bottom up approach), as is the current definition, legal units should be combined into (global) Enterprise Groups. These Enterprise Groups would then be divided into Enterprises based on their autonomy within the group (top down approach). The proposed new definition of Enterprise Group is: The Enterprise group is a set of legal units under common direct or indirect control. It mostly appears as a combination of legal units bound together by legal and/or financial links. The proposed new definition of Enterprise is: The Enterprise is a single legal unit (including a natural person), an Enterprise group as a set of legal units under common control, or a part of an Enterprise group, producing goods or services. It benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current resources. The Enterprise will mostly appear as an organisational unit that can provide meaningful data for statistics. The Truncated Enterprise Group is that part of the Enterprise Group located in a particular country, and the truncated Enterprise is the part of the global Enterprise that belongs to the truncated Enterprise group. The ESSnet on Profiling s preferred method for implementing these new definitions is to profile global groups, divide the global group into Global Enterprises, then into national Truncated Enterprises. Then send the Enterprises list of Legal Units to NSIs for registers & data collection. Implementing this approach requires international communication and co-operation. The aim is to implement using a prototyping approach and gradually. It was mentioned in discussion that The Dutch profiling team consists of 6 staff members who deal with the top 350 groups and a further 15 who deal with the next 2,000 groups. There are different intensities of profiling applied depending on the size of the Enterprise. The French expert described the work of the Profiling Group in INSEE. The profiling department had 4 staff until recently but has now been expanded to 9 staff. He presented two specific examples of how an Enterprise groups restructuring led to an apparent (but not real) sudden change in the statistics. In the case of a major company s restructuring, From one day to the next: the company s indicators production, sale ) doubled and the number of legal units of the sector highly increased, while in reality, nothing changed! In a second example, a company s restructuring gave the impression of a decrease in Industrial turnover, and a corresponding increase in turnover in the wholesale sector, while again nothing had really changed. The most important targets of the profiling group are the 89 groups in France that have 10,000 employees or more, making up 22% of total employees, and 26% of total value added. Profiling is done with the cooperation of the group, and there is an annual individual interview to update the group s profile. The Irish expert described the work of the Large Cases Unit(LCU) in the Irish statistical office. The LCU acts as the single point of contact for the respondents and passes clean data to the survey areas. It deals with 50 to 60 Enterprise groups and obtains data from a wide range of sources, including the tax authorities. He emphasized that Multinational groups constantly change their structures or activities and on going profiling is essential.